apathy

Sutton & Barto is the Bible for RL. Having it be freely available updated is wonderful!

Anyone who hasn't visited Cambodia (or SE Asia in general) should go. At least once in your life. It's cheap, the people are kind (do not insult the king of Thailand though) and the incredible diversity of people streaming in from India, China, and later France has left behind an amazing cultural chronicle. Cambodia will, for better or worse, forcibly remind you of both the heights (Angkor) and the depths (the Khmer Rouge) of humanity.

Note that the KR almost certainly would not have come to power and killed 1/3rd of the population if we hadn't bombed the shit out of eastern Cambodia during the Vietnam war. (Henry Kissinger received the Nobel peace prize for these antics) Almost exactly like Isis, except the KR didn't see the point of destroying their own heritage. So in spite of the odd bullet hole, Angkor and surrounding sites are as awesome as they were when Mouhot stumbled upon them in ~1855. (The Frenchman could not believe that "savages" built such incredible temples and attributed them to the Romans.)

Amazingly resilient people with a glorious legacy that continues to grow as we learn more about it.

I started out as a chemist (undergrad), became a statistician (grad), and finally realized that I had to design, run, analyze and fund my own experiments if I wanted something resembling "the truth", 9 times out of 10.

But my undergrad English teacher made me realize that it's more powerful to plant the seed of an idea and water it with evidence (selectively presented or not) than to try and outright convince people, when you have the luxury of time. I hope that my students are smarter for it; I let them sink or swim by their own wits and I expect them to ask for help if they find they're out of their depth. It's how the real world tends to work anyways.

I'm kind of pissed that I guy I disagreed with seems to have been banned here, for example. I was curious where he was coming from and whether my foundation was as shaky as his appeared to me. (I think it was a guy, but I'll never know)

The best teachers I've ever had simply laid out all the evidence they deemed relevant and had us make up our own minds. The conclusion was never really in doubt by the teachers, but the process of arriving at it independently was more important than the destination.

Now I'm a little curious what happens to blood insulin when athletes fire down a stevia-flavored source of short chain carbohydrates prior to exercise vs. unflavored. When I have time I'll look it up. The nice thing about this (stevia vs. sugar or vs. artificial sweeteners) is that it lends itself naturally to a two factor design, "natural" vs. not and "real sugar" vs. not, orthogonally.

Anyways, yes, I was wrong about that and will probably fuck it up again in the future, although at least now I'm aware of the problem.

You logic strikes me as specious -- the proximity of an idea to the mushy middle is orthogonal to whether it is correct. Great ideas are often met with violent opposition -- but shitty ideas are, too. If I thought most people on HN were in fact dumbasses, as most of the electorate often seems to be, I would go somewhere else. It's the least worst forum (save perhaps Metafilter) I've found so it's a place where I go to procrastinate.

My job is to write, so my natural impulse when I see something that seems easily dismantled is to get in a bit of practice.

Part of the reason for this is that there was no surplus of cheap calories and many people died of trauma or infection prior to becoming susceptible to degenerative disease.

As recently as 1900, the primary cause of death in the United States was infectious diseases. All-cause age-adjusted mortality in the population has halved since then. I suppose we could reduce the role of degenerative and "Western" diseases rather dramatically by discontinuing antibiotics and sanitation, but recent trips to Cambodia and Zimbabwe have me wondering if that's really what you want.

> our tools are finally sharp enough to our liking (fMRI).

fMRI is a false positive generating machine. I've never seen shittier statistical analyses than I see in "neuroscience" (at least the mouse guys use controls).

http://care.diabetesjournals.org/content/early/2013/04/30/dc...

Beware of small sample sizes in these sorts of studies. Until you get up into the hundreds or thousands, I personally would expect regression to the mean. Just FYI.

Why does this not surprise me?

Outside of mathematics, there really isn't anything resembling proof. There is the weight of the evidence, and the weight of any substantive critiques of bias in said evidence. That's about it.

> But there are many research articles out there that look impressive because of all the tools they throw at the problem. Under the surface these studies are bunk.

On this we can certainly agree. However, the replacement for a bad study is a well-designed study, not some statistical hocus pocus (1) or "going by your gut".

If your position is sound, there will (at some point) be someone bold and independently-funded enough to torpedo the status quo with a properly designed experiment. It may not happen as fast as you'd like. But there is nothing that academics like better than turning their rivals' sacred cows into hamburger. Of this I can assure you.

(1) I am a statistician, both professionally and by graduate training. I have no qualms about dismissing poorly designed studies regardless of post-hoc jiggery-pokery. A shiny autopsy won't reanimate the corpse of a dead experiment.

I'm going to go out on a limb here and propose that you may not be a professional medicinal chemist.

Many theories conveniently support your position. Of course, many theories conveniently support the other side. How can we objectively determine who is right?

Well, let's look at (among many, many other experimental studies) http://www.ncbi.nlm.nih.gov/pubmed/25786106 which is a recent study of metabolic differences between self-reported aspartame sensitive and non-sensitive individuals. Double-blind, randomised crossover study -- the gold standard. Any differences in metabolic profile between either control/aspartame or sensitive/not? None seen.

However, perhaps all non-nutritive sweeteners are not alike. The same senior author as above ran a separate study in a small sample of athletic males where aspartame in addition to carbohydrates resulted in lower insulin response during exercise: https://jissn.biomedcentral.com/articles/10.1186/1550-2783-9...

Meanwhile, and in a different (grossly obese, NNS-naive) population, sucralose (Stevia) appeared to do the opposite: http://care.diabetesjournals.org/content/early/2013/04/30/dc...

> Many of these questions can be answered without slow, hard science

Sure, if we don't care whether the answers are credible.

Proof by analogy is fraud. If we didn't need evidence to make decisions we wouldn't bother accumulating it.

Honest-to-god shamans that are the recipients of generations' worth of accumulated cultural traditions aren't my concern; I'm hardly fit to judge them. But that doesn't seem to be what's going on in some of these ceremonies. It's irresponsible (and kind of shitty) to act as a guide unless the guide is prepared to handle extremal behavior.

As far as judging others for taking hallucinogens, nah. People's brains react differently to different agents at different points in time, and that's something that you either learn to deal with or you stop taking hallucinogens.

Reporting incorrect data can be worse than reporting nothing at all. Here are the guidelines: http://www.cdc.gov/nhsn/pdfs/faqs/faqs-v-6_clabsi.pdf

CLIs are only supposed to be reported when the central line can reasonably be implicated as the vector and the blood is the primary infection site. Neither was established.

As far as cleaning up procedures, there will always be room for further improvement, but that's no reason to accept a $2 million fine (which still had to be paid) as a cost of doing business when the "root cause" was incorrect reporting.

Actually I have all sorts of examples of the above, but the point is that oftentimes the data is "corrupted" on the way to the collection point(s). Anyways, the executive (VP of some bullshit) got his walking papers and funnily enough, no more reportable line infections that didn't come from lines.

So at least sometimes there are happy endings, sort of...

Turns out Bostock explicitly avoided devising Yet Another Grammar of Graphics in favor of selection and transformation operators. Which, as it happens, is a very sparse but expressive grammar anyways (cf dplyr and other corners of the Hadleyverse). Nonetheless, I misrepresented his idea of maximizing expressivity by minimizing grammatical constraints.

The IEEE paper is even better than I remembered. It's just as good today as it was when he first submitted it.

As far as the "old" crowd -- who gets sick most often?! It certainly isn't strapping young men and women, not in any hospital I've ever been in outside of a war zone.

Part of the problem is the sheer volume of such cases -- where did the infection come from? Why did it get worse so fast? Was it bacterial, viral, fungal? The diversity of microbial life on earth is so vast, and the rate of evolution so quick, that any one center is going to be hard pressed to keep up just cataloging new strains, let alone figuring out what sequence of drugs to try.

This is why it is a national tragedy (at least in the US) that we don't put more emphasis on funding the CDC. Other countries do better or worse, and particularly some states like Minnesota do a lot more. But I can reasonably guarantee that the possibility of humanity being devastated by an easily transmitted bacterial or viral vector is greater than anything a YF-22A is likely to ward off. A centralized repository and clearinghouse can do so much more than any one center, even the richest hospitals, because it can (in principle) see everything at every center. Just needs enough funding and oversight to get the job done.

I recognize that nothing can bring your MIL back, but perhaps if the CDC were given a bit more respect (instead of being a political football to defund abortion and legalize Confederate flags, as at the present), you might have some answers and the next person in that situation might pull through. No guarantees in life, but it seems like a higher expected return on investment than many government priorities. JMHO.

Sorry for your loss, and the open questions around it.

But if you're claiming to "guide" people through a trip with potential for physical harm (and in an inustrialized, urban environment, that's pretty much all of them), you're goddamn right I'll judge you if you fuck someone up.

If a hospital pulled this shit and didn't even try to mitigate potential harm, they'd be shut down. Would you claim it should be otherwise? People certainly have out of body experiences in the ER on a fairly regular basis...

So, no, I don't take issue with people using hallucinogens. I used a ton when I was younger and had a vanishingly small number of bad trips. The part I take issue with is where people get irresponsible. Humans are pretty fragile (stop by the ER some time, they can survive a lot, but not everything). I judge those who won't take care of each other.

LSD is like walking through the doors of perception.

DMT (as in ayahuasca) is like being shot out of a cannon.

I wonder if it has dawned on any of these "shamans" that a fistful of risperidone might be useful for the freakouts? Antipsychotics exist for a reason. Also, anyone on MAOIs is likely to have a very bad time, and people on SSRIs may not notice a thing (other than barfing).

The whole enterprise seems very poorly thought out.

No dispute that Wilkinson's ideas deeply influenced all sane graphics frameworks thereafter. In fact SPSS' purchase of SYSTAT was (by far) the best thing they ever did, and that's why people still use that awful piece of garbage.

However, Hadley Wickham spent a lot of time thinking through which parts still made sense, and which didn't, in implementing ggplot2. That book is free online, as are Bostock's notes.

I heartily recommend all three, but if money is an issue you can get the latter two for a lot less than a copy of TGoG.

Funfact: Wilkinson started this by selling SYSTAT to SPSS, got bored, eventually went to Tableau (seems to have worked out nicely for them) and now is at h2o. As much as I despise SPSS from having taught classes that use it, I admire their business sense and wish I'd owned equity in them prior to IBM buying the company.

H2O might not be a bad investment either (I've been wrong plenty of times though).

Clearly this library is setting the world on fire. Meanwhile if you want slow pointy clicky chartjunk for your PHB you can probably use your company's Tableau license, and if you want to push the envelope you use d3.

I've been using d3 on and off since 2011. It started from a spare, elegant idea and progressed to the point that almost anything I might wish to implement has at least some pieces already written in some obscure blocks.org block.

I find that pretty damned impressive given that many of my interests were research projects abandoned in the 90s and only just now beginning to come back. (yes, there were smart people back then, too; they all seemed to have SGIs)

Unlike Zika tests, it is extremely rare for patients in the trials I design to have any alternatives. Also unlike Zika, the standard of care in the diseases I work on yields less than 25% 5-year survival.

Not birth defects (horrific though they may be) but death from organ failure and infection. Nearly all of the compounds are known or potential teratogens. No alternative sources for most of the drugs. Still need an IRB.

You say, "oh it's just a finger stick!" To which I must say, have you looked at the number of physicians and nurses who got Hep C or HIV from an errant finger stick in the hospital?!

Common sense in medicine prioritizes doing no harm.

I can provide other notable examples if need be. The bottom line is that all parties need to be confident that the expected benefit of a test or procedure outweighs any expected potential harms. Having multiple layers or peer review is one way to ensure this.

I work on clinical trials of "black label" (potentially lethal) agents in patients that typically have no other options. Our pages can be 70-100 pages and we are doing this in our "spare" time (Ha! Ha!). I don't think it's unreasonable for a massively capitalized private company to produce documentation at least up to this standard.

It really isn't much to ask, especially for an existential venture like this (something -- anything! -- trustworthy to recover the tremendous amount of credibility they've lost.)

Other companies had no problem writing a suitable protocol. Why should peoples' safety be compromised just so that an untrustworthy company can enter the market?

IRBs exist to ensure that these sorts of things are not overlooked. Too many times in the past, they have been. So now the burden of proof lies with the investigator.

Like it or not, a few fast-and-loose protocols ensured that this situation will not be changing any time soon.